Smoking indicators such as breath carbon monoxide (CO) and semiquantitative cotinine immunoassay test strips (urine and saliva NicAlert) are frequently relied on to determine smoking status, but few studies have evaluated the relative performance of these indicators alone and in combination. Additionally, there is still disagreement about the optimal breath CO cutoff to discriminate smokers from nonsmokers. Nontreatment seeking smokers (heavy and light) and nonsmokers (exposed and not exposed to passive smoking environments) completed smoking histories and provided breath CO, urine, and saliva specimens. Urine and saliva specimens were assayed for cotinine by NicAlert and liquid chromatography-tandem mass spectrometry (LCMSMS). An optimal breath CO cutoff was established using self-report and gold standard LCMSMS analysis of cotinine as reference measures. Performance of self-report, breath CO (at optimal cutoff), urine NicAlert, saliva NicAlert, and combinations of these indicators were compared to LCMSMS as the reference. Breath CO &#8805;5 ppm optimally discriminated smokers from nonsmokers. Saliva NicAlert performance was less effective than the other indicators, which performed similarly in predicting smoking status determined by LCMSMS. The optimal breath CO cutoff of &#8805;5 ppm is lower than the &#8805;8-10 ppm cutoff recommended by the Society for Research on Nicotine and Tobacco in 2002. Recent public health initiatives to decrease passive smoke exposure may have lowered breath CO from environmental exposure, allowing a lower cutoff level to efficiently differentiate smokers from nonsmokers. A combination of self-report, breath CO (cutoff &#8805;5 ppm), and urine NicAlert testing can optimally determine smoking status.